Introduction

The model for telecom service delivery is changing in response to advances in internet technologies. Historically, the architecture of a telecom service delivery platform was based upon large, expensive equipment and a monolithic clustered design with many tightly integrated components. Communication technologies were integrated tightly with the applications. This approach resisted change at many levels.

 

Advances in service delivery brought about by the internet have dramatically changed many of the concepts for modern service delivery. In the internet, networking technologies are largely separated from application design, allowing application designers to focus on what they do best. Hardware is now deployed in modular, low cost increments using a "rack and stack" server farm methodology. This approach enables a "grow as you go" concept with low cost, independent units replicated for redundancy, performance, and dynamic scalability when needed.

 

To reduce complexity, functions tend to be grouped in architectural segments. Those functions that require clustering, for example a database, tend to be isolated from the main application functional groups. Clustering remains a useful solution for the special needs of some communication and database services but often simply adds cost without corresponding value for the application.

 

Also, routing components can offload critical communications functions from application server farm elements by providing capabilities such as load balancing, optimized message steering, address conversions, and overload management. We can apply the lessons of the internet to telecom applications and service delivery.

 

Challenges

• Apply modern IT server farm technologies to telecom network application deployments
• Reduce service application development/support complexity (and cost)
• Enable cost-effective, finely-tuned performance scalability
• Maintain carrier-grade high availability model for telecom core network services
• Provide SS7/SIGTRAN network capabilities for services built and deployed in server farms
• Create a system solution that embraces change

Many challenges face the designers of new telecom services and those wishing to integrate internet capabilities into telecom networks. We must take advantage of new technology solutions. We must integrate "made-for-IP" services with traditional telecom network capabilities. There exists a need for a way to leverage the cost advantages of state-of-the-art equipment technologies and service development/deployment techniques.

 

In particular, you might ask: how do you take an existing service that conformed to a traditional telecom service model, and deploy it (with minimal change) on a server farm to leverage the cost advantages of new technologies? How can you reduce complexity so as to reduce both initial costs and ongoing maintenance costs? How can you introduce change to the service dynamically without service interruption? How can you introduce access to traditional carrier services into existing internet applications? How do you meet "carrier-grade" expectations for high availability using the internet services model and server farm technology while, at the same time, reducing the complexity of the solution?

 

Solution

In order to meet these challenges, we must address the following issues:

• Hide the internal structure of the server farm from the signaling network so that service can be dynamically (re)structuredwhile running. This also creates a security layer that hides internal information from those who have no need for it
• Using the internet-model, treat each instance of the service application/ farm element as non-redundant, but provide redundancy through replicated server farm elements that are managed to provide carrier-grade high availability to the service user
• Outboard functions from the application, such as load distribution, high availability, and congestion management, to a signaling gateway so that service applications do not have to deal with these complexities. Use the gateway to manage the traffic distribution across the available farm elements, keeping in mind their dynamic characteristics such as instantaneous load or congestion as well as considering their relative capacity/ performance characteristics
• Enable service scaling by simply adding new farm elements - where the hardware and software of these service components are simply "cloned" as needed. Allow mixtures of older, slower components with newer, faster components to preserve investments as you go
• Isolate the application from network considerations. Use a signaling gateway to provide the many international variants of signaling protocols and physical interfaces while isolating the application servers from these considerations
• Manage sets of related application service instances so that the network sees a single logical service rather than many instances of a service. Provide management proxies for the service that take into consideration how many instances are necessary to reliably offer a service
• Protect service applications from bad network behavior using screening services and traffic burst management
• Provide service optimization mechanisms such as content-based message steering to optimize service application database caches or transaction context management

Ulticom nSignia eSTP

Ulticom's nSignia^® eSTP addresses these challenges by providing the above solutions in an elegant, turnkey network element that meets the highest demands for performance and reliability. nSignia eSTP enables the creation of service applications that leverage the benefits of modern server farm technologies while being able to operate in both the IP-based internet space and the traditional SS7-based telecom signaling networks. nSignia eSTP brings carrier-grade high availability to internet style applications. Many of the features of an IP router are provided by nSignia eSTP so that service applications can be designed to address their problem space in much the same way that internet applications approach their problem space. But nSignia eSTP goes further, providing capabilities such as multi-component availability management and overload control mechanisms that have yet to be provided in IP routers. Thus, nSignia eSTP is a true telecom solution, combining best-of-breed capabilities from both the internet and the traditional carrier-grade services space.

 

The capabilities of nSignia eSTP can significantly reduce application and deployment complexity by enabling the internet server farm model for deployment. Reduced Complexity = Reduced Cost.

 

Key Benefits

Improved overall system cost/performance ratio - nSignia eSTP allows the application to be deployed on few or many low cost standard server farm computing elements, with the ability to "grow as you go". nSignia eSTP represents this collection of application servers to the core network as a single entity using proxy technologies, and enables the network to access the farm without the need to understand its internal makeup or handle component failures or component upgrades

Lower cost of initial service deployment - using a server farm approach, initial deployments can be right-sized, without fear of impact to the ongoing service operation when the need for capacity upgrade is required. nSignia eSTP provides many of the critical functions necessary for carrier-grade service deployment, so the application designers do not have to address these costly considerations in their application design

Use of best-of-breed server farms using nSignia eSTP and server farm technologies, service providers can chose modern internet-honed equipment and management tools. This can dramatically lower cost and increase usability and manageability

Reduced ongoing software support costs - nSignia eSTP provides much of the non-revenue generating functionality traditionally added to network service applications. By keeping this functionality out of the application, the total size of the application is reduced, and the costs of creating and repairing this aspect of the functionality are removed from the overall support structure

Information hiding for increased application security - nSignia eSTP hides the details of how many farm elements provide a service. The network addresses the nSignia eSTP, not the application servers. No information about the number of application servers or which ones provide a particular service is visible. Security is becoming increasingly important as the IP and traditional carrier networks collide. nSignia eSTP can also screen out messages from an unfriendly network element to further aid application security

Service application software and equipment (re)configuration without service interruption - by hiding the internal configuration of a server farm from the core network, and by managing the load distribution across the server farm, nSignia eSTP can also hide individual component state and upgrades from the core network. Failures, addition of new farm elements, and hardware or software upgrades of farm elements are all invisible to the core network

Dynamic, efficient load distribution and message steering to minimize transaction response times and optimize equipment utilization - nSignia eSTP provides mechanisms to inspect message content beyond header level and can use this to steer transactions to specific farm elements to enable optimized handling of transactions and customers

Optimized service scaling using the internet-model - the internet scaling model is based upon the concept of adding new farm elements as the need exists. Using nSignia eSTP, this model is supported for telecom, carrier-grade applications as well

Creates a natural approach to enable the introduction of virtualization into the telecom service applications space which, in turn, confers additional cost and management benefits - by hiding the structure of the server farm from the core network, we have the additional opportunity to rearrange the structure of the service, while it is running. In addition to adding or deleting hardware components, we can reorganize then using virtual machine technologies to create optimized deployment strategies. This can be particularly useful when addressing small initial operational capability environments in which multiple components can be combined on a single physical instance to minimize initial deployment costs, but later expand as the service utilization grows